| 摘要 |
In an ATM network using a burst-level band-width allocation, a source terminal reserves, before a burst transmission, band-widths according to a peak rate of the burst and then sends the burst therethrough. When the transmission is finished, the band-widths are released. Where there exists a link to which the peak rate is not assigned, NACK is sent to the terminal and the reserved band-widths are released. On receiving the NACK, the terminal allocates a band-width with peak rate lower than that of the first request after a back-off time is elapsed, thereby minimizing the block probability of the transmission. The source terminal declares a minimum band-width together with the peak rate (maximum band-width) in the band-width request operation. Each node allocates the peak rate when the remaining band-width of a link controlled by the node is sufficient to allocate the peak rate. Even if the remaining band-width is insufficient, when the band-width is not less than the minimum band-width, there is allocated a band-width equal to or more than the minimum band-width and equal to or less than the peak rate according to the remaining band-width, thereby transferring the burst. <IMAGE> |
